MAPS (MIT Array Performance Simulator)

Haystack Observatory has been working on developing concepts, designs and tools for low frequency interferometric arrays for the past many years. These arrays will operate far beyond the routine operating regime of most existing interferometers. They will have to contend with imaging wide fields of view in presence of intense galactic background emission, severe ionospheric corruption of the incoming astronomical signal and RFI. The task of high dynamic range imaging will be further complicated by the presence of variable beam shapes effects. In order to drive the development of calibration and imaging algorithms for such arrays, the Haystack group is developing the MIT Array Performance Simulator (MAPS), a simulator work package which produces interferometric data-sets, realistically incorporating the various effects mentioned above. MAPS can accommodate the most detailed description of heterogeneous interferometric arrays, down to individual receptor level, allows station based beam forming and arbitrary descriptions of a time and location dependent ionosphere above the array and an arbitrary sky brightness distribution over the entire sky. The primary purpose of MAPS is to simulate realistic synthetic data sets to test the array performance and proposed calibration and imaging methods.

Current Murchison Widefield Array (MWA) designs plan on achieving near noise limited images using an innovative, but largely untested, calibration scheme. This simulator work package will also be used to explore the applicability of present day imaging algorithms to MWA data and serve as a starting point for development of new algorithms. A refined simulation capability will also be a powerful tool for searching through MWA design parameter space to help determine performance tradeoffs.Though MAPS was originally developed in context of the low frequency array projects at Haystack, it is a very general and versatile tool which can be used for simulating data from any interferometer. The fact that it is currently in use to assess the performance of SKA candidate arrays by the SKA Simulations Working Group is a fitting demonstration of its flexibility. A recent version of MAPS is now available for use via web pages maintained at Swinburne University.

Requirements:
In order to perform realistic and detailed simulations for an interferometer of arbitrary description MAPS must include all the effects which may corrupt the data and taken the following approach.

Array Geometry: Array geometry can be specified in the simulator down to placement and orientation of the individual receptors. The receptors are described by individual beam patterns. This allows the possibility of using dipole arrays as well as standard dishes. MAPS can combine individual receptors to generate beam-formed station beams, which realistically change as a function of time and frequency. It will also let users explore the effects of dipole damage, and optimize dipole placement within a station. On larger scales, station layout can be specified on any 2-D grid which is then mapped onto a curved Earth surface for accurate calculation of (u,v,w) values. Receptor patterns for all receptors used in the array can be independently specified. This should allow one to simulate any conceivable interferometric array.
Input Brightness Distribution: Elliptical Gaussian and Point sources can be specified in an easily readable text file. These are sampled onto a brightness array (truncated at a low level to keep computation times to a minimum). Binary images imported from AIPS, AIPS++, or general image editor can be added to the brightness array. This allows real scientific images or benchmark test patterns to serve as observational targets for the simulator. Sources outside the field of view that may enter via sidelobes can also be included.
Observing Specifications: All major parameters governing a 'virtual' observation are specified in a text file. In the frequency domain, one can specify individual channel widths and frequencies. Multiple observation start and stop pairs are supported. Field of view center and size and correlator integration time are also set in this file.
Thermal Noise: A Gaussian noise term can be added to the complex visibilities. This will depend on the number of dipoles in a station and on the sky they are sensitive to. Maps of the Galactic Background will be used to calculate the sky contribution to the thermal noise as a function of time.
Variable Station Beams: At low frequencies, the interferometer elements are likely to consist of a large number (~64) receptors phased together to form a station beam. The shape and gain of the station beam will naturally change with time as the station beam formers compensate for the rotation of the Earth. Due to the large fractional bandwidths involved, beams at lower frequencies may be quite different from beams at higher frequencies. Because the receptor positions are known, the simulator can calculate station beams for each correlator integration time and frequency channel and for each station. Each complex visibility corresponding to a (u,v,w) point and specific baseline will include the effects of the beam shapes from each contributing station.
Ionosphere: Calibration of the ionosphere represents one of the major challenges to low frequency array design and the simulator must be capable of including effects of a realistic ionosphere. At the heart of this part of the simulator is a 4 dimensional (x,y,z,t) ionospheric model that represents known behaviors of the ionosphere. For each integration time and for each station, 2-D phase screens will be constructed which correspond to the integrated TEC over the field of view. These screens will evolve with time and scale with frequency and they will be used to calculate the visibilities on each baseline separately.
Export: Output from the simulator is exported directly into AIPS++ where familiar FITS files can be produced for import into Classic AIPS, or most any other interferometry reduction software.

Some memos of particular interest: (These memos were generated as part of MIT's LOFAR effort.)

SEARCH: